AI in Facility Management - PPM

?From the Editor.

AI has become the biggest ‘buzzword’ in the last couple of years with claims that it can replicate almost any task that humans can perform.

AI or NOT AI that is the question?

Real AI platforms such as ChatGPT cost a fortune to make with OpenAI's losses growing to $540 million last year ?and could cost 700k per day to use given the costs involved with computing power according to reports.

So how can so many companies claim their software / sensors are AI driven especially those in the CMMS / EAM sectors?

It is confusing for prospective buyers where they are being offered ‘Star Trek’ and being given ‘Flash Gordan’.

The reality is that most vendors have only improved the programming in their software but full AI – No.

In this newsletter we will look at the potential of AI in PPM and is it currently a viable proposition.

Dominic Murnane CEO

CompassAFM

?The human body is the most amazing result of countless years of evolution. I remember the words of my instructor when I started out 40 years ago as a mechanical engineering apprentice. He said “Look at the top of your finger, the technology behind what it can do will take engineers another 200 years before they can completely replicate it”. And he was right!

AI works by processing vast amounts of data, learning from patterns within it, and using this knowledge to make predictions, decisions, or perform tasks without explicit programming.

Can this work in facilities management and can it replace human knowledge and intuition? To get an insight to this I ‘interviewed’ ChatGPT.

Dominic

Can AI work in facilities PPM?

ChatGPT

Yes, AI can be highly beneficial in facilities preventive maintenance (PPM). AI algorithms can analyse historical maintenance data to predict equipment failures before they occur, allowing for proactive maintenance scheduling. Additionally, AI can monitor equipment in real-time, detecting anomalies and triggering alerts for potential issues. This predictive maintenance approach helps prevent costly breakdowns and extends the lifespan of equipment. AI can also optimize maintenance schedules based on factors like equipment usage patterns and environmental conditions, ensuring maintenance is performed when it's most efficient and least disruptive. Overall, integrating AI into facilities PPM can lead to increased efficiency, reduced costs, and improved facility performance.

Dominic

When analysing historical maintenance data to predict equipment failures, where are you obtaining this data?

ChatGPT

The data can be obtained from various sources such as maintenance logs, work orders, equipment sensors, and maintenance management systems (CMMS).

?Dominic

If maintenance logs / reports are handwritten what do you do then?

ChatGPT

If maintenance logs or reports are handwritten, they need to be digitized first…

Dominic

So unless everything is digitised, that data source is unavailable?

ChatGPT

That's correct. Unless handwritten maintenance logs or reports are digitized, the data within them remains inaccessible for automated analysis by AI systems.

Dominic

In the case of an expensive plant item such as an Air Handling Unit, how are sensors going to predict a failure?

ChatGPT

Sensors can predict failures by monitoring various parameters such as temperature, pressure, vibration, and energy consumption. Anomalies or deviations from normal operating conditions detected by these sensors can indicate potential issues. Additionally, sensors can track performance trends over time, identifying gradual degradation that could lead to failure if left unaddressed.

Dominic

So can sensors identify structural damage such as damaged hinges, rust? Can they tell if filters have been incorrectly installed? Can they tell that a belt is wearing?

ChatGPT

Sensors can detect various types of structural damage and wear in equipment like Air Handling Units (AHUs), including damaged hinges, rust, incorrectly installed filters, and belt wear. For instance:

1. Structural Damage and Rust: Sensors can monitor vibrations and acoustic signals, detecting anomalies that may indicate structural issues or corrosion. Additionally, visual sensors or cameras can be used to inspect physical conditions and identify rust or damage.
2. Incorrectly Installed Filters: Pressure sensors can monitor airflow and pressure differentials across filters. If a filter is incorrectly installed or clogged, it can cause changes in these parameters, triggering alerts.
3. Belt Wear: Sensors can monitor belt tension, vibration patterns, and motor current draw to assess the condition of belts. As belts wear, tension changes and vibrations may increase, indicating the need for maintenance.

While sensors alone might not directly identify these issues as humans would, AI algorithms can interpret sensor data to recognize patterns indicative of these problems.

Dominic

What would be an estimate of the costs of fitting all possible sensors including cameras for visual defects that AI could monitor for an AHU?

ChatGPT

For a typical AHU system, you might need several sensors of each type to adequately monitor various parameters. Let's say you want to monitor temperature, humidity, pressure, vibration, and airflow, and you install two sensors of each type. Additionally, you decide to include two visual inspection cameras.

Estimated costs:

Temperature and Humidity Sensors (2): $100 - $400

Pressure Sensors (2): $200 - $1000

Vibration Sensors (2): $200 - $1000

Current Sensors (2): $100 - $400

Airflow Sensors (2): $200 - $1000

Visual Inspection Cameras (2): $400 - $2000

Data Transmission and Communication Equipment: $500 - $2000

?Installation Costs: $5000 - $10,000

Total estimated cost range: $6600 - $22,800

Dominic

And the cost of the AI system to monitor them?

ChatGPT

The cost of the AI system to monitor sensors in an Air Handling Unit (AHU) can vary widely depending on several factors, including the complexity of the AI algorithms, the level of customization required, and whether the system is developed in-house or purchased from a vendor.

Let's provide a ballpark estimate assuming you opt for a third-party AI solution:

• Initial Development/Setup Costs: $10,000 - $50,000 (depending on customization)
• Annual Licensing/Subscription Fees: $5,000 - $20,000
• Cloud Computing Costs: $5,000 - $20,000 per year (depending on usage)
• Maintenance and Support: $2,000 - $10,000 per year

Total estimated cost range for the AI system: $22,000 - $100,000+ per year

So, let’s take a reality check. Equipping plant with enough sensors to detect as much potential faults as possible could easily equal 1/5 of the plant value. The typical lifespan of these sensors is 5 years, so they will most likely have to be replaced at least once during the lifespan of the plant. And then you have the cost of the AI system monitoring the sensors.

The plant still must be maintained regularly, and a skilled technician’s abilities will still be far superior in detecting potential issues during the maintenance. While some sensors connect to a BMS have undoubted valuable in detecting faults, can a full AI powered sensor system be worthwhile? I don’t think so.

So, let’s continue with the interview.

Dominic

When analysing data input in a CMMS for example a Mitsubishi AC Ceiling Cassette System, what maintenance data would you require?

ChatGPT (Abridged)

Several types of maintenance data are crucial for effective management and decision-making. Here's a breakdown of the key maintenance data required:

1. Maintenance Activities: (Date and time of maintenance, Type of maintenance (e.g., PPM).
2. Maintenance Personnel: Information about the individuals or teams performing
3. Work Orders: Records of past work orders generated for maintenance tasks
4. Fault Reports: Reports of any faults or malfunctions observed or reported for the AC system
5. Equipment Downtime: Duration and frequency of equipment downtime
6. Parts and Materials Used: Inventory of parts and materials used during maintenance tasks
7. Maintenance Costs: Costs associated with maintenance activities
8. Condition Monitoring Data: Data from condition monitoring sensors installed on the AC system
9. Scheduled Maintenance Plans: Planned maintenance schedules for the AC system, including use of consumable parts (e.g., filters).
10. Historical Maintenance Records: Historical data on past maintenance activities.

Dominic

This data is primarily administrative, what about the data from a checklist of tasks needed to be performed during maintenance?

ChatGPT (Abridged)

You're absolutely right. In addition to administrative data, maintenance data should include detailed information from checklists outlining tasks to be performed during maintenance. Here's the breakdown of maintenance data from checklists…

?Dominic

The checklists you generated were accurate and would cover the majority of required tasks. How would you propose to obtain and analyse this data from the technician?

ChatGPT (Abridged)

To obtain and analyse maintenance data from technicians, several methods can be used to ensure accurate and efficient data collection and analysis:

·?????? Digital Checklists: Provide technicians with digital checklists accessible through a mobile app or tablet. Technicians can fill out the checklist electronically during maintenance activities, recording observations, measurements, and task completion status in real-time.

·?????? Data Validation and Quality Checks: Implement validation rules to ensure data accuracy and consistency. This includes checks for required fields, valid data ranges, and logical relationships between data points. Technicians can be prompted to correct any errors before submitting the checklist.

·?????? Analytical Tools: Utilize analytical tools within the CMMS or integrated software to analyse maintenance data collected from technicians. These tools can generate reports, charts, and dashboards to visualize trends, identify recurring issues, and track key performance indicators (KPIs) related to maintenance activities.

·?????? Machine Learning and AI: Incorporate machine learning and AI algorithms to analyse maintenance data and identify patterns or anomalies. These algorithms can automatically flag potential issues, predict maintenance needs, and recommend optimization strategies based on historical data.

Dominic

So again maintenance data must be digitised and input to the CMMS

ChatGPT

Yes, digitizing maintenance data and inputting it into a CMMS (Computerized Maintenance Management System) offers several benefits for effective maintenance management and analysis.

Dominic

A technician has access to specific maintenance manuals and manufacturing specifications for the plant they are maintaining. How does an AI have this access and be able to use this information to improve its analysis?

ChatGPT

Estimating the cost for an AI system to access and utilize information from maintenance manuals and specifications involves several factors. Here's a breakdown of potential costs:

1. Digitization of Manuals and Specifications: This involves scanning and converting paper-based manuals into digital formats, as well as organizing and storing them in a centralized repository. Costs can range from $5,000 to $20,000, depending on the volume of documents and complexity.
2. Text Recognition (OCR) Technology: Implementing OCR technology to extract text from digital documents. Costs can vary based on the sophistication of the OCR solution and the number of documents to process, ranging from $5,000 to $20,000.
3. Knowledge Graph Development: Creating knowledge graphs to represent relationships between components, procedures, and specifications. Costs depend on the complexity of the graphs and can range from $10,000 to $50,000.
4. Natural Language Processing (NLP) Models: Developing NLP models to analyze and understand maintenance manuals. Costs for NLP model development can vary widely but may range from $20,000 to $100,000.
5. Machine Learning Models: Training machine learning models to recognize patterns and extract relevant information from manuals. Costs for ML model development can range from $30,000 to $150,000, depending on complexity.
6. Semantic Analysis Tools: Implementing semantic analysis tools to understand the meaning and context of information. Costs can vary but may range from $10,000 to $50,000.
7. Rule-based Systems: Developing rule-based systems to encode domain-specific knowledge. Costs can range from $10,000 to $50,000.
8. Continuous Learning Infrastructure: Setting up infrastructure for continuous learning, including data storage, model retraining, and updates. Costs may range from $10,000 to $50,000.

Adding these costs together, the total estimated cost for implementing an AI system to access and utilize information from maintenance manuals and specifications could range from $100,000 to $440,000 for the first year. Ongoing costs for maintenance, updates, and improvements may add an additional $20,000 to $100,000 per year.

Dominic

What if your CMMS cannot handle this level of data input?

?ChatGPT

If your CMMS cannot handle the level of data input required you need to

1. Evaluate CMMS Capabilities: Assess your CMMS to determine its limitations and identify which aspects of data input it struggles with. Is it the interface, the storage capacity, the reporting capabilities, or something else?
2. Communicate with CMMS Provider: Reach out to your CMMS provider to discuss your requirements and see if there are any upgrades or additional modules available that could better meet your needs. They might offer solutions tailored to your specific requirements.
3. Explore Alternative Solutions: If your CMMS is severely limited and cannot be adequately adapted, consider exploring alternative CMMS options that better align with your requirements. Look for CMMS systems with robust data input capabilities and scalability.

Dominic

Besides the AI cost you mentioned to monitor sensors above, what costs are we looking at for an AI that can interrogate and analyse data from a CMMS?

ChatGPT

Certainly, let's provide a rough estimate of the costs involved in implementing an AI system to interrogate and analyse data from a CMMS:

1. Data Integration Costs: $5,000 - $20,000
2. Data Preprocessing: $2,000 - $10,000
3. Hardware and Infrastructure: $10,000 - $50,000
4. AI Development and Training: $50,000 - $200,000
5. Software Licensing and Tools: $5,000 - $20,000
6. Testing and Validation: $10,000 - $30,000
7. Deployment and Integration: $10,000 - $50,000
8. Training and Support: $5,000 - $20,000
9. Regulatory Compliance: $5,000 - $30,000
10. Updates and Maintenance: $10,000 - $50,000 per year

Total estimated cost range for implementing AI to interrogate and analyze CMMS data: $112,000 - $480,000 for the first year, with ongoing costs of $10,000 - $50,000 per year thereafter.

Dominic

If an organisation had 1,000 plant items from 40 vendors and all were to be equipped with sensors and all maintenance data digitised in a CMMS. Based on what we discussed above and including staff training how long would you estimate it would take an AI system to be fully operational?

ChatGPT (Abridged)

Based on these estimates, the total time to implement and operationalize the AI system could range from 2 to 3.5 years. This timeline accounts for the complexity of integrating data from a large number of plant items, developing AI models, and ensuring the system is fully functional and adopted by staff.

Dominic

From what we discussed for the 1000 plant items, sensor installation, CMMS data digitalisation and full AI implementation. What would you estimate is the installation cost and recurring annual costs?

ChatGPT (Abridged)

Estimated Total Installation Cost: $675,000 - $925,000

Estimated Total Recurring Annual Costs: $210,000

Please note that these are rough estimates and actual costs may vary based on factors such as the chosen vendors, the complexity of the AI system, and specific requirements of the organization. Additionally, these estimates do not include potential costs for data storage, additional software licenses, or unforeseen expenses.

Dominic

For the size of organisation we discussed above, what is the possibility of a CMMS vendor offering a software package claiming full AI capability and offering it for under $150,000 per year performing all the functions we have discussed?

ChatGPT

It's not possible that a CMMS vendor could offer a software package claiming full AI capability for $150,000 per year, encompassing the functions we discussed.

?Conclusion

Is it possible to buy a functional AI system for your facilities maintenance – Yes, if you are prepared to invest huge sums of money and time.

Will is perform as well as skilled contractors and FM team - No?

At the moment it would be more practical, cost effective and beneficial to employ a couple of highly skilled facility managers to perform analytics on all the data input from remedial and maintenance works as well as having a good CMMS such as FacilityManager 6 that can easily handle the data input and provide good analytical tools.

FacilityManager 6

Elegance, Simplicity and Power in a CMMS

CompassAFM customers are using real-time diagnostic data from their plant maintenance to reduce downtime, cut maintenance costs and make their plant more efficient. “Since launching FacilityManager, it has been a huge success,” explains Dominic Murnane CEO. “We saw the potential for how it could be a tremendous benefit to our clients.” Typically, our clients have greater that 4X ROI.

If you do not have a CMMS solution or your current CMMS is outdated or not fit for purpose, then FacilityManager? from CompassAFM is the perfect answer.

FacilityManager? is a multi-award winning CMMS software solution designed to be intuitive, exceptionally easy to use and with all the features you will need and more. We have clients who are high end users with just a few hours training.

FacilityManager? allows you to efficiently manage your Planned / Preventative Maintenance (Plant & Building), Work Orders and much more backed by detailed analytics and reporting.

I can’t think of any good reason why you wouldn’t want to explore how FacilityManager? can benefit your organisation.

What’s New in v 6?

·?????? Contract interior cleaning module

·?????? Smart scanning & labelling

·?????? Daily Task Explorer

·?????? Improved navigation and searching

·?????? Improved layouts for Android mobile devices

·?????? Risk Assessment & Staff Training Module

·?????? User Login Tracker

·?????? New Contractor Module

·?????? Re-designed KPI’s

·?????? New Asset & Property Registers

·?????? 3rd Party Fault Receipt & Ticket Generation

·?????? Staff Training Record

·?????? New Calendar / Planner

·?????? New Contractor Gateway w Contractor Induction

·?????? Compass Training Academy

·?????? Integration with BMS and other monitoring systems

Why not schedule a demo or activate a free 90 day, unlimited use trial, simply use the link below.

No commitment. No obligation. No card details required. Trust us, you won't regret it!

FacilityManager Demo / Trial?

Upcoming Events

??? 3 Day Facility Management Masterclass FREE, from 18:00pm to 19:00pm each day over Zoom

??? 3 Day Work Safety Masterclass FREE, from 09:00am to 10:00am each day over Zoom

??? CompassASSIST – How to transform your FM 1 Day FREE, from 09:00am to 12:00am over Zoom

??? FacilityManager CMMS Q&A 1 Day, from 09:00am to 10:00am over Zoom

If you would like to get more information on the above article, or our upcoming events -

just write “AI or Events” in the comments.

Arrange a 1 – 1 call via Calendly.

https://calendly.com/compassafm/30-min-demo

Follow me for more interesting articles -?https://lnkd.in/erpZNyCB

Email:?????????????? [email protected]

WhatsApp: ?????? 00353 87 6076565

Phone / Text: ?? 00353 87 6076565